4 TGSI, Tungsten Graphics Shader Infrastructure, is an intermediate language
5 for describing shaders. Since Gallium is inherently shaderful, shaders are
6 an important part of the API. TGSI is the only intermediate representation
12 All TGSI instructions, known as *opcodes*, operate on arbitrary-precision
13 floating-point four-component vectors. An opcode may have up to one
14 destination register, known as *dst*, and between zero and three source
15 registers, called *src0* through *src2*, or simply *src* if there is only
18 Some instructions, like :opcode:`I2F`, permit re-interpretation of vector
19 components as integers. Other instructions permit using registers as
20 two-component vectors with double precision; see :ref:`doubleopcodes`.
22 When an instruction has a scalar result, the result is usually copied into
23 each of the components of *dst*. When this happens, the result is said to be
24 *replicated* to *dst*. :opcode:`RCP` is one such instruction.
29 TGSI supports modifiers on inputs (as well as saturate modifier on instructions).
31 For inputs which have a floating point type, both absolute value and negation
32 modifiers are supported (with absolute value being applied first).
33 TGSI_OPCODE_MOV is considered to have float input type for applying modifiers.
35 For inputs which have signed or unsigned type only the negate modifier is
42 ^^^^^^^^^^^^^^^^^^^^^^^^^
44 These opcodes are guaranteed to be available regardless of the driver being
47 .. opcode:: ARL - Address Register Load
51 dst.x = \lfloor src.x\rfloor
53 dst.y = \lfloor src.y\rfloor
55 dst.z = \lfloor src.z\rfloor
57 dst.w = \lfloor src.w\rfloor
60 .. opcode:: MOV - Move
73 .. opcode:: LIT - Light Coefficients
78 dst.y &= max(src.x, 0) \\
79 dst.z &= (src.x > 0) ? max(src.y, 0)^{clamp(src.w, -128, 128))} : 0 \\
83 .. opcode:: RCP - Reciprocal
85 This instruction replicates its result.
92 .. opcode:: RSQ - Reciprocal Square Root
94 This instruction replicates its result. The results are undefined for src <= 0.
98 dst = \frac{1}{\sqrt{src.x}}
101 .. opcode:: SQRT - Square Root
103 This instruction replicates its result. The results are undefined for src < 0.
110 .. opcode:: EXP - Approximate Exponential Base 2
114 dst.x &= 2^{\lfloor src.x\rfloor} \\
115 dst.y &= src.x - \lfloor src.x\rfloor \\
116 dst.z &= 2^{src.x} \\
120 .. opcode:: LOG - Approximate Logarithm Base 2
124 dst.x &= \lfloor\log_2{|src.x|}\rfloor \\
125 dst.y &= \frac{|src.x|}{2^{\lfloor\log_2{|src.x|}\rfloor}} \\
126 dst.z &= \log_2{|src.x|} \\
130 .. opcode:: MUL - Multiply
134 dst.x = src0.x \times src1.x
136 dst.y = src0.y \times src1.y
138 dst.z = src0.z \times src1.z
140 dst.w = src0.w \times src1.w
143 .. opcode:: ADD - Add
147 dst.x = src0.x + src1.x
149 dst.y = src0.y + src1.y
151 dst.z = src0.z + src1.z
153 dst.w = src0.w + src1.w
156 .. opcode:: DP3 - 3-component Dot Product
158 This instruction replicates its result.
162 dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z
165 .. opcode:: DP4 - 4-component Dot Product
167 This instruction replicates its result.
171 dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z + src0.w \times src1.w
174 .. opcode:: DST - Distance Vector
179 dst.y &= src0.y \times src1.y\\
184 .. opcode:: MIN - Minimum
188 dst.x = min(src0.x, src1.x)
190 dst.y = min(src0.y, src1.y)
192 dst.z = min(src0.z, src1.z)
194 dst.w = min(src0.w, src1.w)
197 .. opcode:: MAX - Maximum
201 dst.x = max(src0.x, src1.x)
203 dst.y = max(src0.y, src1.y)
205 dst.z = max(src0.z, src1.z)
207 dst.w = max(src0.w, src1.w)
210 .. opcode:: SLT - Set On Less Than
214 dst.x = (src0.x < src1.x) ? 1.0F : 0.0F
216 dst.y = (src0.y < src1.y) ? 1.0F : 0.0F
218 dst.z = (src0.z < src1.z) ? 1.0F : 0.0F
220 dst.w = (src0.w < src1.w) ? 1.0F : 0.0F
223 .. opcode:: SGE - Set On Greater Equal Than
227 dst.x = (src0.x >= src1.x) ? 1.0F : 0.0F
229 dst.y = (src0.y >= src1.y) ? 1.0F : 0.0F
231 dst.z = (src0.z >= src1.z) ? 1.0F : 0.0F
233 dst.w = (src0.w >= src1.w) ? 1.0F : 0.0F
236 .. opcode:: MAD - Multiply And Add
240 dst.x = src0.x \times src1.x + src2.x
242 dst.y = src0.y \times src1.y + src2.y
244 dst.z = src0.z \times src1.z + src2.z
246 dst.w = src0.w \times src1.w + src2.w
249 .. opcode:: SUB - Subtract
253 dst.x = src0.x - src1.x
255 dst.y = src0.y - src1.y
257 dst.z = src0.z - src1.z
259 dst.w = src0.w - src1.w
262 .. opcode:: LRP - Linear Interpolate
266 dst.x = src0.x \times src1.x + (1 - src0.x) \times src2.x
268 dst.y = src0.y \times src1.y + (1 - src0.y) \times src2.y
270 dst.z = src0.z \times src1.z + (1 - src0.z) \times src2.z
272 dst.w = src0.w \times src1.w + (1 - src0.w) \times src2.w
275 .. opcode:: CND - Condition
279 dst.x = (src2.x > 0.5) ? src0.x : src1.x
281 dst.y = (src2.y > 0.5) ? src0.y : src1.y
283 dst.z = (src2.z > 0.5) ? src0.z : src1.z
285 dst.w = (src2.w > 0.5) ? src0.w : src1.w
288 .. opcode:: DP2A - 2-component Dot Product And Add
292 dst.x = src0.x \times src1.x + src0.y \times src1.y + src2.x
294 dst.y = src0.x \times src1.x + src0.y \times src1.y + src2.x
296 dst.z = src0.x \times src1.x + src0.y \times src1.y + src2.x
298 dst.w = src0.x \times src1.x + src0.y \times src1.y + src2.x
301 .. opcode:: FRC - Fraction
305 dst.x = src.x - \lfloor src.x\rfloor
307 dst.y = src.y - \lfloor src.y\rfloor
309 dst.z = src.z - \lfloor src.z\rfloor
311 dst.w = src.w - \lfloor src.w\rfloor
314 .. opcode:: CLAMP - Clamp
318 dst.x = clamp(src0.x, src1.x, src2.x)
320 dst.y = clamp(src0.y, src1.y, src2.y)
322 dst.z = clamp(src0.z, src1.z, src2.z)
324 dst.w = clamp(src0.w, src1.w, src2.w)
327 .. opcode:: FLR - Floor
329 This is identical to :opcode:`ARL`.
333 dst.x = \lfloor src.x\rfloor
335 dst.y = \lfloor src.y\rfloor
337 dst.z = \lfloor src.z\rfloor
339 dst.w = \lfloor src.w\rfloor
342 .. opcode:: ROUND - Round
355 .. opcode:: EX2 - Exponential Base 2
357 This instruction replicates its result.
364 .. opcode:: LG2 - Logarithm Base 2
366 This instruction replicates its result.
373 .. opcode:: POW - Power
375 This instruction replicates its result.
379 dst = src0.x^{src1.x}
381 .. opcode:: XPD - Cross Product
385 dst.x = src0.y \times src1.z - src1.y \times src0.z
387 dst.y = src0.z \times src1.x - src1.z \times src0.x
389 dst.z = src0.x \times src1.y - src1.x \times src0.y
394 .. opcode:: ABS - Absolute
407 .. opcode:: RCC - Reciprocal Clamped
409 This instruction replicates its result.
411 XXX cleanup on aisle three
415 dst = (1 / src.x) > 0 ? clamp(1 / src.x, 5.42101e-020, 1.84467e+019) : clamp(1 / src.x, -1.84467e+019, -5.42101e-020)
418 .. opcode:: DPH - Homogeneous Dot Product
420 This instruction replicates its result.
424 dst = src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z + src1.w
427 .. opcode:: COS - Cosine
429 This instruction replicates its result.
436 .. opcode:: DDX, DDX_FINE - Derivative Relative To X
438 The fine variant is only used when ``PIPE_CAP_TGSI_FS_FINE_DERIVATIVE`` is
439 advertised. When it is, the fine version guarantees one derivative per row
440 while DDX is allowed to be the same for the entire 2x2 quad.
444 dst.x = partialx(src.x)
446 dst.y = partialx(src.y)
448 dst.z = partialx(src.z)
450 dst.w = partialx(src.w)
453 .. opcode:: DDY, DDY_FINE - Derivative Relative To Y
455 The fine variant is only used when ``PIPE_CAP_TGSI_FS_FINE_DERIVATIVE`` is
456 advertised. When it is, the fine version guarantees one derivative per column
457 while DDY is allowed to be the same for the entire 2x2 quad.
461 dst.x = partialy(src.x)
463 dst.y = partialy(src.y)
465 dst.z = partialy(src.z)
467 dst.w = partialy(src.w)
470 .. opcode:: PK2H - Pack Two 16-bit Floats
475 .. opcode:: PK2US - Pack Two Unsigned 16-bit Scalars
480 .. opcode:: PK4B - Pack Four Signed 8-bit Scalars
485 .. opcode:: PK4UB - Pack Four Unsigned 8-bit Scalars
490 .. opcode:: RFL - Reflection Vector
494 dst.x = 2 \times (src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z) / (src0.x \times src0.x + src0.y \times src0.y + src0.z \times src0.z) \times src0.x - src1.x
496 dst.y = 2 \times (src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z) / (src0.x \times src0.x + src0.y \times src0.y + src0.z \times src0.z) \times src0.y - src1.y
498 dst.z = 2 \times (src0.x \times src1.x + src0.y \times src1.y + src0.z \times src1.z) / (src0.x \times src0.x + src0.y \times src0.y + src0.z \times src0.z) \times src0.z - src1.z
504 Considered for removal.
507 .. opcode:: SEQ - Set On Equal
511 dst.x = (src0.x == src1.x) ? 1.0F : 0.0F
513 dst.y = (src0.y == src1.y) ? 1.0F : 0.0F
515 dst.z = (src0.z == src1.z) ? 1.0F : 0.0F
517 dst.w = (src0.w == src1.w) ? 1.0F : 0.0F
520 .. opcode:: SFL - Set On False
522 This instruction replicates its result.
530 Considered for removal.
533 .. opcode:: SGT - Set On Greater Than
537 dst.x = (src0.x > src1.x) ? 1.0F : 0.0F
539 dst.y = (src0.y > src1.y) ? 1.0F : 0.0F
541 dst.z = (src0.z > src1.z) ? 1.0F : 0.0F
543 dst.w = (src0.w > src1.w) ? 1.0F : 0.0F
546 .. opcode:: SIN - Sine
548 This instruction replicates its result.
555 .. opcode:: SLE - Set On Less Equal Than
559 dst.x = (src0.x <= src1.x) ? 1.0F : 0.0F
561 dst.y = (src0.y <= src1.y) ? 1.0F : 0.0F
563 dst.z = (src0.z <= src1.z) ? 1.0F : 0.0F
565 dst.w = (src0.w <= src1.w) ? 1.0F : 0.0F
568 .. opcode:: SNE - Set On Not Equal
572 dst.x = (src0.x != src1.x) ? 1.0F : 0.0F
574 dst.y = (src0.y != src1.y) ? 1.0F : 0.0F
576 dst.z = (src0.z != src1.z) ? 1.0F : 0.0F
578 dst.w = (src0.w != src1.w) ? 1.0F : 0.0F
581 .. opcode:: STR - Set On True
583 This instruction replicates its result.
590 .. opcode:: TEX - Texture Lookup
592 for array textures src0.y contains the slice for 1D,
593 and src0.z contain the slice for 2D.
595 for shadow textures with no arrays (and not cube map),
596 src0.z contains the reference value.
598 for shadow textures with arrays, src0.z contains
599 the reference value for 1D arrays, and src0.w contains
600 the reference value for 2D arrays and cube maps.
602 for cube map array shadow textures, the reference value
603 cannot be passed in src0.w, and TEX2 must be used instead.
609 shadow_ref = src0.z or src0.w (optional)
613 dst = texture\_sample(unit, coord, shadow_ref)
616 .. opcode:: TEX2 - Texture Lookup (for shadow cube map arrays only)
618 this is the same as TEX, but uses another reg to encode the
629 dst = texture\_sample(unit, coord, shadow_ref)
634 .. opcode:: TXD - Texture Lookup with Derivatives
646 dst = texture\_sample\_deriv(unit, coord, ddx, ddy)
649 .. opcode:: TXP - Projective Texture Lookup
653 coord.x = src0.x / src0.w
655 coord.y = src0.y / src0.w
657 coord.z = src0.z / src0.w
663 dst = texture\_sample(unit, coord)
666 .. opcode:: UP2H - Unpack Two 16-Bit Floats
672 Considered for removal.
674 .. opcode:: UP2US - Unpack Two Unsigned 16-Bit Scalars
680 Considered for removal.
682 .. opcode:: UP4B - Unpack Four Signed 8-Bit Values
688 Considered for removal.
690 .. opcode:: UP4UB - Unpack Four Unsigned 8-Bit Scalars
696 Considered for removal.
698 .. opcode:: X2D - 2D Coordinate Transformation
702 dst.x = src0.x + src1.x \times src2.x + src1.y \times src2.y
704 dst.y = src0.y + src1.x \times src2.z + src1.y \times src2.w
706 dst.z = src0.x + src1.x \times src2.x + src1.y \times src2.y
708 dst.w = src0.y + src1.x \times src2.z + src1.y \times src2.w
712 Considered for removal.
715 .. opcode:: ARA - Address Register Add
721 Considered for removal.
723 .. opcode:: ARR - Address Register Load With Round
736 .. opcode:: SSG - Set Sign
740 dst.x = (src.x > 0) ? 1 : (src.x < 0) ? -1 : 0
742 dst.y = (src.y > 0) ? 1 : (src.y < 0) ? -1 : 0
744 dst.z = (src.z > 0) ? 1 : (src.z < 0) ? -1 : 0
746 dst.w = (src.w > 0) ? 1 : (src.w < 0) ? -1 : 0
749 .. opcode:: CMP - Compare
753 dst.x = (src0.x < 0) ? src1.x : src2.x
755 dst.y = (src0.y < 0) ? src1.y : src2.y
757 dst.z = (src0.z < 0) ? src1.z : src2.z
759 dst.w = (src0.w < 0) ? src1.w : src2.w
762 .. opcode:: KILL_IF - Conditional Discard
764 Conditional discard. Allowed in fragment shaders only.
768 if (src.x < 0 || src.y < 0 || src.z < 0 || src.w < 0)
773 .. opcode:: KILL - Discard
775 Unconditional discard. Allowed in fragment shaders only.
778 .. opcode:: SCS - Sine Cosine
791 .. opcode:: TXB - Texture Lookup With Bias
793 for cube map array textures and shadow cube maps, the bias value
794 cannot be passed in src0.w, and TXB2 must be used instead.
796 if the target is a shadow texture, the reference value is always
797 in src.z (this prevents shadow 3d and shadow 2d arrays from
798 using this instruction, but this is not needed).
814 dst = texture\_sample(unit, coord, bias)
817 .. opcode:: TXB2 - Texture Lookup With Bias (some cube maps only)
819 this is the same as TXB, but uses another reg to encode the
820 lod bias value for cube map arrays and shadow cube maps.
821 Presumably shadow 2d arrays and shadow 3d targets could use
822 this encoding too, but this is not legal.
824 shadow cube map arrays are neither possible nor required.
834 dst = texture\_sample(unit, coord, bias)
837 .. opcode:: NRM - 3-component Vector Normalise
841 dst.x = src.x / (src.x \times src.x + src.y \times src.y + src.z \times src.z)
843 dst.y = src.y / (src.x \times src.x + src.y \times src.y + src.z \times src.z)
845 dst.z = src.z / (src.x \times src.x + src.y \times src.y + src.z \times src.z)
850 .. opcode:: DIV - Divide
854 dst.x = \frac{src0.x}{src1.x}
856 dst.y = \frac{src0.y}{src1.y}
858 dst.z = \frac{src0.z}{src1.z}
860 dst.w = \frac{src0.w}{src1.w}
863 .. opcode:: DP2 - 2-component Dot Product
865 This instruction replicates its result.
869 dst = src0.x \times src1.x + src0.y \times src1.y
872 .. opcode:: TXL - Texture Lookup With explicit LOD
874 for cube map array textures, the explicit lod value
875 cannot be passed in src0.w, and TXL2 must be used instead.
877 if the target is a shadow texture, the reference value is always
878 in src.z (this prevents shadow 3d / 2d array / cube targets from
879 using this instruction, but this is not needed).
895 dst = texture\_sample(unit, coord, lod)
898 .. opcode:: TXL2 - Texture Lookup With explicit LOD (for cube map arrays only)
900 this is the same as TXL, but uses another reg to encode the
902 Presumably shadow 3d / 2d array / cube targets could use
903 this encoding too, but this is not legal.
905 shadow cube map arrays are neither possible nor required.
915 dst = texture\_sample(unit, coord, lod)
918 .. opcode:: PUSHA - Push Address Register On Stack
927 Considered for cleanup.
931 Considered for removal.
933 .. opcode:: POPA - Pop Address Register From Stack
942 Considered for cleanup.
946 Considered for removal.
949 .. opcode:: BRA - Branch
955 Considered for removal.
958 .. opcode:: CALLNZ - Subroutine Call If Not Zero
964 Considered for cleanup.
968 Considered for removal.
972 ^^^^^^^^^^^^^^^^^^^^^^^^
974 These opcodes are primarily provided for special-use computational shaders.
975 Support for these opcodes indicated by a special pipe capability bit (TBD).
977 XXX doesn't look like most of the opcodes really belong here.
979 .. opcode:: CEIL - Ceiling
983 dst.x = \lceil src.x\rceil
985 dst.y = \lceil src.y\rceil
987 dst.z = \lceil src.z\rceil
989 dst.w = \lceil src.w\rceil
992 .. opcode:: TRUNC - Truncate
1000 dst.z = trunc(src.z)
1002 dst.w = trunc(src.w)
1005 .. opcode:: MOD - Modulus
1009 dst.x = src0.x \bmod src1.x
1011 dst.y = src0.y \bmod src1.y
1013 dst.z = src0.z \bmod src1.z
1015 dst.w = src0.w \bmod src1.w
1018 .. opcode:: UARL - Integer Address Register Load
1020 Moves the contents of the source register, assumed to be an integer, into the
1021 destination register, which is assumed to be an address (ADDR) register.
1024 .. opcode:: SAD - Sum Of Absolute Differences
1028 dst.x = |src0.x - src1.x| + src2.x
1030 dst.y = |src0.y - src1.y| + src2.y
1032 dst.z = |src0.z - src1.z| + src2.z
1034 dst.w = |src0.w - src1.w| + src2.w
1037 .. opcode:: TXF - Texel Fetch
1039 As per NV_gpu_shader4, extract a single texel from a specified texture
1040 image. The source sampler may not be a CUBE or SHADOW. src 0 is a
1041 four-component signed integer vector used to identify the single texel
1042 accessed. 3 components + level. Just like texture instructions, an optional
1043 offset vector is provided, which is subject to various driver restrictions
1044 (regarding range, source of offsets).
1045 TXF(uint_vec coord, int_vec offset).
1048 .. opcode:: TXQ - Texture Size Query
1050 As per NV_gpu_program4, retrieve the dimensions of the texture depending on
1051 the target. For 1D (width), 2D/RECT/CUBE (width, height), 3D (width, height,
1052 depth), 1D array (width, layers), 2D array (width, height, layers).
1053 Also return the number of accessible levels (last_level - first_level + 1)
1056 For components which don't return a resource dimension, their value
1064 dst.x = texture\_width(unit, lod)
1066 dst.y = texture\_height(unit, lod)
1068 dst.z = texture\_depth(unit, lod)
1070 dst.w = texture\_levels(unit)
1072 .. opcode:: TG4 - Texture Gather
1074 As per ARB_texture_gather, gathers the four texels to be used in a bi-linear
1075 filtering operation and packs them into a single register. Only works with
1076 2D, 2D array, cubemaps, and cubemaps arrays. For 2D textures, only the
1077 addressing modes of the sampler and the top level of any mip pyramid are
1078 used. Set W to zero. It behaves like the TEX instruction, but a filtered
1079 sample is not generated. The four samples that contribute to filtering are
1080 placed into xyzw in clockwise order, starting with the (u,v) texture
1081 coordinate delta at the following locations (-, +), (+, +), (+, -), (-, -),
1082 where the magnitude of the deltas are half a texel.
1084 PIPE_CAP_TEXTURE_SM5 enhances this instruction to support shadow per-sample
1085 depth compares, single component selection, and a non-constant offset. It
1086 doesn't allow support for the GL independent offset to get i0,j0. This would
1087 require another CAP is hw can do it natively. For now we lower that before
1096 dst = texture\_gather4 (unit, coord, component)
1098 (with SM5 - cube array shadow)
1106 dst = texture\_gather (uint, coord, compare)
1108 .. opcode:: LODQ - level of detail query
1110 Compute the LOD information that the texture pipe would use to access the
1111 texture. The Y component contains the computed LOD lambda_prime. The X
1112 component contains the LOD that will be accessed, based on min/max lod's
1119 dst.xy = lodq(uint, coord);
1122 ^^^^^^^^^^^^^^^^^^^^^^^^
1123 These opcodes are used for integer operations.
1124 Support for these opcodes indicated by PIPE_SHADER_CAP_INTEGERS (all of them?)
1127 .. opcode:: I2F - Signed Integer To Float
1129 Rounding is unspecified (round to nearest even suggested).
1133 dst.x = (float) src.x
1135 dst.y = (float) src.y
1137 dst.z = (float) src.z
1139 dst.w = (float) src.w
1142 .. opcode:: U2F - Unsigned Integer To Float
1144 Rounding is unspecified (round to nearest even suggested).
1148 dst.x = (float) src.x
1150 dst.y = (float) src.y
1152 dst.z = (float) src.z
1154 dst.w = (float) src.w
1157 .. opcode:: F2I - Float to Signed Integer
1159 Rounding is towards zero (truncate).
1160 Values outside signed range (including NaNs) produce undefined results.
1173 .. opcode:: F2U - Float to Unsigned Integer
1175 Rounding is towards zero (truncate).
1176 Values outside unsigned range (including NaNs) produce undefined results.
1180 dst.x = (unsigned) src.x
1182 dst.y = (unsigned) src.y
1184 dst.z = (unsigned) src.z
1186 dst.w = (unsigned) src.w
1189 .. opcode:: UADD - Integer Add
1191 This instruction works the same for signed and unsigned integers.
1192 The low 32bit of the result is returned.
1196 dst.x = src0.x + src1.x
1198 dst.y = src0.y + src1.y
1200 dst.z = src0.z + src1.z
1202 dst.w = src0.w + src1.w
1205 .. opcode:: UMAD - Integer Multiply And Add
1207 This instruction works the same for signed and unsigned integers.
1208 The multiplication returns the low 32bit (as does the result itself).
1212 dst.x = src0.x \times src1.x + src2.x
1214 dst.y = src0.y \times src1.y + src2.y
1216 dst.z = src0.z \times src1.z + src2.z
1218 dst.w = src0.w \times src1.w + src2.w
1221 .. opcode:: UMUL - Integer Multiply
1223 This instruction works the same for signed and unsigned integers.
1224 The low 32bit of the result is returned.
1228 dst.x = src0.x \times src1.x
1230 dst.y = src0.y \times src1.y
1232 dst.z = src0.z \times src1.z
1234 dst.w = src0.w \times src1.w
1237 .. opcode:: IMUL_HI - Signed Integer Multiply High Bits
1239 The high 32bits of the multiplication of 2 signed integers are returned.
1243 dst.x = (src0.x \times src1.x) >> 32
1245 dst.y = (src0.y \times src1.y) >> 32
1247 dst.z = (src0.z \times src1.z) >> 32
1249 dst.w = (src0.w \times src1.w) >> 32
1252 .. opcode:: UMUL_HI - Unsigned Integer Multiply High Bits
1254 The high 32bits of the multiplication of 2 unsigned integers are returned.
1258 dst.x = (src0.x \times src1.x) >> 32
1260 dst.y = (src0.y \times src1.y) >> 32
1262 dst.z = (src0.z \times src1.z) >> 32
1264 dst.w = (src0.w \times src1.w) >> 32
1267 .. opcode:: IDIV - Signed Integer Division
1269 TBD: behavior for division by zero.
1273 dst.x = src0.x \ src1.x
1275 dst.y = src0.y \ src1.y
1277 dst.z = src0.z \ src1.z
1279 dst.w = src0.w \ src1.w
1282 .. opcode:: UDIV - Unsigned Integer Division
1284 For division by zero, 0xffffffff is returned.
1288 dst.x = src0.x \ src1.x
1290 dst.y = src0.y \ src1.y
1292 dst.z = src0.z \ src1.z
1294 dst.w = src0.w \ src1.w
1297 .. opcode:: UMOD - Unsigned Integer Remainder
1299 If second arg is zero, 0xffffffff is returned.
1303 dst.x = src0.x \ src1.x
1305 dst.y = src0.y \ src1.y
1307 dst.z = src0.z \ src1.z
1309 dst.w = src0.w \ src1.w
1312 .. opcode:: NOT - Bitwise Not
1325 .. opcode:: AND - Bitwise And
1329 dst.x = src0.x \& src1.x
1331 dst.y = src0.y \& src1.y
1333 dst.z = src0.z \& src1.z
1335 dst.w = src0.w \& src1.w
1338 .. opcode:: OR - Bitwise Or
1342 dst.x = src0.x | src1.x
1344 dst.y = src0.y | src1.y
1346 dst.z = src0.z | src1.z
1348 dst.w = src0.w | src1.w
1351 .. opcode:: XOR - Bitwise Xor
1355 dst.x = src0.x \oplus src1.x
1357 dst.y = src0.y \oplus src1.y
1359 dst.z = src0.z \oplus src1.z
1361 dst.w = src0.w \oplus src1.w
1364 .. opcode:: IMAX - Maximum of Signed Integers
1368 dst.x = max(src0.x, src1.x)
1370 dst.y = max(src0.y, src1.y)
1372 dst.z = max(src0.z, src1.z)
1374 dst.w = max(src0.w, src1.w)
1377 .. opcode:: UMAX - Maximum of Unsigned Integers
1381 dst.x = max(src0.x, src1.x)
1383 dst.y = max(src0.y, src1.y)
1385 dst.z = max(src0.z, src1.z)
1387 dst.w = max(src0.w, src1.w)
1390 .. opcode:: IMIN - Minimum of Signed Integers
1394 dst.x = min(src0.x, src1.x)
1396 dst.y = min(src0.y, src1.y)
1398 dst.z = min(src0.z, src1.z)
1400 dst.w = min(src0.w, src1.w)
1403 .. opcode:: UMIN - Minimum of Unsigned Integers
1407 dst.x = min(src0.x, src1.x)
1409 dst.y = min(src0.y, src1.y)
1411 dst.z = min(src0.z, src1.z)
1413 dst.w = min(src0.w, src1.w)
1416 .. opcode:: SHL - Shift Left
1418 The shift count is masked with 0x1f before the shift is applied.
1422 dst.x = src0.x << (0x1f \& src1.x)
1424 dst.y = src0.y << (0x1f \& src1.y)
1426 dst.z = src0.z << (0x1f \& src1.z)
1428 dst.w = src0.w << (0x1f \& src1.w)
1431 .. opcode:: ISHR - Arithmetic Shift Right (of Signed Integer)
1433 The shift count is masked with 0x1f before the shift is applied.
1437 dst.x = src0.x >> (0x1f \& src1.x)
1439 dst.y = src0.y >> (0x1f \& src1.y)
1441 dst.z = src0.z >> (0x1f \& src1.z)
1443 dst.w = src0.w >> (0x1f \& src1.w)
1446 .. opcode:: USHR - Logical Shift Right
1448 The shift count is masked with 0x1f before the shift is applied.
1452 dst.x = src0.x >> (unsigned) (0x1f \& src1.x)
1454 dst.y = src0.y >> (unsigned) (0x1f \& src1.y)
1456 dst.z = src0.z >> (unsigned) (0x1f \& src1.z)
1458 dst.w = src0.w >> (unsigned) (0x1f \& src1.w)
1461 .. opcode:: UCMP - Integer Conditional Move
1465 dst.x = src0.x ? src1.x : src2.x
1467 dst.y = src0.y ? src1.y : src2.y
1469 dst.z = src0.z ? src1.z : src2.z
1471 dst.w = src0.w ? src1.w : src2.w
1475 .. opcode:: ISSG - Integer Set Sign
1479 dst.x = (src0.x < 0) ? -1 : (src0.x > 0) ? 1 : 0
1481 dst.y = (src0.y < 0) ? -1 : (src0.y > 0) ? 1 : 0
1483 dst.z = (src0.z < 0) ? -1 : (src0.z > 0) ? 1 : 0
1485 dst.w = (src0.w < 0) ? -1 : (src0.w > 0) ? 1 : 0
1489 .. opcode:: FSLT - Float Set On Less Than (ordered)
1491 Same comparison as SLT but returns integer instead of 1.0/0.0 float
1495 dst.x = (src0.x < src1.x) ? \sim 0 : 0
1497 dst.y = (src0.y < src1.y) ? \sim 0 : 0
1499 dst.z = (src0.z < src1.z) ? \sim 0 : 0
1501 dst.w = (src0.w < src1.w) ? \sim 0 : 0
1504 .. opcode:: ISLT - Signed Integer Set On Less Than
1508 dst.x = (src0.x < src1.x) ? \sim 0 : 0
1510 dst.y = (src0.y < src1.y) ? \sim 0 : 0
1512 dst.z = (src0.z < src1.z) ? \sim 0 : 0
1514 dst.w = (src0.w < src1.w) ? \sim 0 : 0
1517 .. opcode:: USLT - Unsigned Integer Set On Less Than
1521 dst.x = (src0.x < src1.x) ? \sim 0 : 0
1523 dst.y = (src0.y < src1.y) ? \sim 0 : 0
1525 dst.z = (src0.z < src1.z) ? \sim 0 : 0
1527 dst.w = (src0.w < src1.w) ? \sim 0 : 0
1530 .. opcode:: FSGE - Float Set On Greater Equal Than (ordered)
1532 Same comparison as SGE but returns integer instead of 1.0/0.0 float
1536 dst.x = (src0.x >= src1.x) ? \sim 0 : 0
1538 dst.y = (src0.y >= src1.y) ? \sim 0 : 0
1540 dst.z = (src0.z >= src1.z) ? \sim 0 : 0
1542 dst.w = (src0.w >= src1.w) ? \sim 0 : 0
1545 .. opcode:: ISGE - Signed Integer Set On Greater Equal Than
1549 dst.x = (src0.x >= src1.x) ? \sim 0 : 0
1551 dst.y = (src0.y >= src1.y) ? \sim 0 : 0
1553 dst.z = (src0.z >= src1.z) ? \sim 0 : 0
1555 dst.w = (src0.w >= src1.w) ? \sim 0 : 0
1558 .. opcode:: USGE - Unsigned Integer Set On Greater Equal Than
1562 dst.x = (src0.x >= src1.x) ? \sim 0 : 0
1564 dst.y = (src0.y >= src1.y) ? \sim 0 : 0
1566 dst.z = (src0.z >= src1.z) ? \sim 0 : 0
1568 dst.w = (src0.w >= src1.w) ? \sim 0 : 0
1571 .. opcode:: FSEQ - Float Set On Equal (ordered)
1573 Same comparison as SEQ but returns integer instead of 1.0/0.0 float
1577 dst.x = (src0.x == src1.x) ? \sim 0 : 0
1579 dst.y = (src0.y == src1.y) ? \sim 0 : 0
1581 dst.z = (src0.z == src1.z) ? \sim 0 : 0
1583 dst.w = (src0.w == src1.w) ? \sim 0 : 0
1586 .. opcode:: USEQ - Integer Set On Equal
1590 dst.x = (src0.x == src1.x) ? \sim 0 : 0
1592 dst.y = (src0.y == src1.y) ? \sim 0 : 0
1594 dst.z = (src0.z == src1.z) ? \sim 0 : 0
1596 dst.w = (src0.w == src1.w) ? \sim 0 : 0
1599 .. opcode:: FSNE - Float Set On Not Equal (unordered)
1601 Same comparison as SNE but returns integer instead of 1.0/0.0 float
1605 dst.x = (src0.x != src1.x) ? \sim 0 : 0
1607 dst.y = (src0.y != src1.y) ? \sim 0 : 0
1609 dst.z = (src0.z != src1.z) ? \sim 0 : 0
1611 dst.w = (src0.w != src1.w) ? \sim 0 : 0
1614 .. opcode:: USNE - Integer Set On Not Equal
1618 dst.x = (src0.x != src1.x) ? \sim 0 : 0
1620 dst.y = (src0.y != src1.y) ? \sim 0 : 0
1622 dst.z = (src0.z != src1.z) ? \sim 0 : 0
1624 dst.w = (src0.w != src1.w) ? \sim 0 : 0
1627 .. opcode:: INEG - Integer Negate
1642 .. opcode:: IABS - Integer Absolute Value
1656 These opcodes are used for bit-level manipulation of integers.
1658 .. opcode:: IBFE - Signed Bitfield Extract
1660 See SM5 instruction of the same name. Extracts a set of bits from the input,
1661 and sign-extends them if the high bit of the extracted window is set.
1665 def ibfe(value, offset, bits):
1666 offset = offset & 0x1f
1668 if bits == 0: return 0
1669 # Note: >> sign-extends
1670 if width + offset < 32:
1671 return (value << (32 - offset - bits)) >> (32 - bits)
1673 return value >> offset
1675 .. opcode:: UBFE - Unsigned Bitfield Extract
1677 See SM5 instruction of the same name. Extracts a set of bits from the input,
1678 without any sign-extension.
1682 def ubfe(value, offset, bits):
1683 offset = offset & 0x1f
1685 if bits == 0: return 0
1686 # Note: >> does not sign-extend
1687 if width + offset < 32:
1688 return (value << (32 - offset - bits)) >> (32 - bits)
1690 return value >> offset
1692 .. opcode:: BFI - Bitfield Insert
1694 See SM5 instruction of the same name. Replaces a bit region of 'base' with
1695 the low bits of 'insert'.
1699 def bfi(base, insert, offset, bits):
1700 offset = offset & 0x1f
1702 mask = ((1 << bits) - 1) << offset
1703 return ((insert << offset) & mask) | (base & ~mask)
1705 .. opcode:: BREV - Bitfield Reverse
1707 See SM5 instruction BFREV. Reverses the bits of the argument.
1709 .. opcode:: POPC - Population Count
1711 See SM5 instruction COUNTBITS. Counts the number of set bits in the argument.
1713 .. opcode:: LSB - Index of lowest set bit
1715 See SM5 instruction FIRSTBIT_LO. Computes the 0-based index of the first set
1716 bit of the argument. Returns -1 if none are set.
1718 .. opcode:: IMSB - Index of highest non-sign bit
1720 See SM5 instruction FIRSTBIT_SHI. Computes the 0-based index of the highest
1721 non-sign bit of the argument (i.e. highest 0 bit for negative numbers,
1722 highest 1 bit for positive numbers). Returns -1 if all bits are the same
1723 (i.e. for inputs 0 and -1).
1725 .. opcode:: UMSB - Index of highest set bit
1727 See SM5 instruction FIRSTBIT_HI. Computes the 0-based index of the highest
1728 set bit of the argument. Returns -1 if none are set.
1731 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
1733 These opcodes are only supported in geometry shaders; they have no meaning
1734 in any other type of shader.
1736 .. opcode:: EMIT - Emit
1738 Generate a new vertex for the current primitive into the specified vertex
1739 stream using the values in the output registers.
1742 .. opcode:: ENDPRIM - End Primitive
1744 Complete the current primitive in the specified vertex stream (consisting of
1745 the emitted vertices), and start a new one.
1751 These opcodes are part of :term:`GLSL`'s opcode set. Support for these
1752 opcodes is determined by a special capability bit, ``GLSL``.
1753 Some require glsl version 1.30 (UIF/BREAKC/SWITCH/CASE/DEFAULT/ENDSWITCH).
1755 .. opcode:: CAL - Subroutine Call
1761 .. opcode:: RET - Subroutine Call Return
1766 .. opcode:: CONT - Continue
1768 Unconditionally moves the point of execution to the instruction after the
1769 last bgnloop. The instruction must appear within a bgnloop/endloop.
1773 Support for CONT is determined by a special capability bit,
1774 ``TGSI_CONT_SUPPORTED``. See :ref:`Screen` for more information.
1777 .. opcode:: BGNLOOP - Begin a Loop
1779 Start a loop. Must have a matching endloop.
1782 .. opcode:: BGNSUB - Begin Subroutine
1784 Starts definition of a subroutine. Must have a matching endsub.
1787 .. opcode:: ENDLOOP - End a Loop
1789 End a loop started with bgnloop.
1792 .. opcode:: ENDSUB - End Subroutine
1794 Ends definition of a subroutine.
1797 .. opcode:: NOP - No Operation
1802 .. opcode:: BRK - Break
1804 Unconditionally moves the point of execution to the instruction after the
1805 next endloop or endswitch. The instruction must appear within a loop/endloop
1806 or switch/endswitch.
1809 .. opcode:: BREAKC - Break Conditional
1811 Conditionally moves the point of execution to the instruction after the
1812 next endloop or endswitch. The instruction must appear within a loop/endloop
1813 or switch/endswitch.
1814 Condition evaluates to true if src0.x != 0 where src0.x is interpreted
1815 as an integer register.
1819 Considered for removal as it's quite inconsistent wrt other opcodes
1820 (could emulate with UIF/BRK/ENDIF).
1823 .. opcode:: IF - Float If
1825 Start an IF ... ELSE .. ENDIF block. Condition evaluates to true if
1829 where src0.x is interpreted as a floating point register.
1832 .. opcode:: UIF - Bitwise If
1834 Start an UIF ... ELSE .. ENDIF block. Condition evaluates to true if
1838 where src0.x is interpreted as an integer register.
1841 .. opcode:: ELSE - Else
1843 Starts an else block, after an IF or UIF statement.
1846 .. opcode:: ENDIF - End If
1848 Ends an IF or UIF block.
1851 .. opcode:: SWITCH - Switch
1853 Starts a C-style switch expression. The switch consists of one or multiple
1854 CASE statements, and at most one DEFAULT statement. Execution of a statement
1855 ends when a BRK is hit, but just like in C falling through to other cases
1856 without a break is allowed. Similarly, DEFAULT label is allowed anywhere not
1857 just as last statement, and fallthrough is allowed into/from it.
1858 CASE src arguments are evaluated at bit level against the SWITCH src argument.
1864 (some instructions here)
1867 (some instructions here)
1870 (some instructions here)
1875 .. opcode:: CASE - Switch case
1877 This represents a switch case label. The src arg must be an integer immediate.
1880 .. opcode:: DEFAULT - Switch default
1882 This represents the default case in the switch, which is taken if no other
1886 .. opcode:: ENDSWITCH - End of switch
1888 Ends a switch expression.
1891 .. opcode:: NRM4 - 4-component Vector Normalise
1893 This instruction replicates its result.
1897 dst = \frac{src.x}{src.x \times src.x + src.y \times src.y + src.z \times src.z + src.w \times src.w}
1903 The interpolation instructions allow an input to be interpolated in a
1904 different way than its declaration. This corresponds to the GLSL 4.00
1905 interpolateAt* functions. The first argument of each of these must come from
1906 ``TGSI_FILE_INPUT``.
1908 .. opcode:: INTERP_CENTROID - Interpolate at the centroid
1910 Interpolates the varying specified by src0 at the centroid
1912 .. opcode:: INTERP_SAMPLE - Interpolate at the specified sample
1914 Interpolates the varying specified by src0 at the sample id specified by
1915 src1.x (interpreted as an integer)
1917 .. opcode:: INTERP_OFFSET - Interpolate at the specified offset
1919 Interpolates the varying specified by src0 at the offset src1.xy from the
1920 pixel center (interpreted as floats)
1928 The double-precision opcodes reinterpret four-component vectors into
1929 two-component vectors with doubled precision in each component.
1931 Support for these opcodes is XXX undecided. :T
1933 .. opcode:: DADD - Add
1937 dst.xy = src0.xy + src1.xy
1939 dst.zw = src0.zw + src1.zw
1942 .. opcode:: DDIV - Divide
1946 dst.xy = src0.xy / src1.xy
1948 dst.zw = src0.zw / src1.zw
1950 .. opcode:: DSEQ - Set on Equal
1954 dst.xy = src0.xy == src1.xy ? 1.0F : 0.0F
1956 dst.zw = src0.zw == src1.zw ? 1.0F : 0.0F
1958 .. opcode:: DSLT - Set on Less than
1962 dst.xy = src0.xy < src1.xy ? 1.0F : 0.0F
1964 dst.zw = src0.zw < src1.zw ? 1.0F : 0.0F
1966 .. opcode:: DFRAC - Fraction
1970 dst.xy = src.xy - \lfloor src.xy\rfloor
1972 dst.zw = src.zw - \lfloor src.zw\rfloor
1975 .. opcode:: DFRACEXP - Convert Number to Fractional and Integral Components
1977 Like the ``frexp()`` routine in many math libraries, this opcode stores the
1978 exponent of its source to ``dst0``, and the significand to ``dst1``, such that
1979 :math:`dst1 \times 2^{dst0} = src` .
1983 dst0.xy = exp(src.xy)
1985 dst1.xy = frac(src.xy)
1987 dst0.zw = exp(src.zw)
1989 dst1.zw = frac(src.zw)
1991 .. opcode:: DLDEXP - Multiply Number by Integral Power of 2
1993 This opcode is the inverse of :opcode:`DFRACEXP`.
1997 dst.xy = src0.xy \times 2^{src1.xy}
1999 dst.zw = src0.zw \times 2^{src1.zw}
2001 .. opcode:: DMIN - Minimum
2005 dst.xy = min(src0.xy, src1.xy)
2007 dst.zw = min(src0.zw, src1.zw)
2009 .. opcode:: DMAX - Maximum
2013 dst.xy = max(src0.xy, src1.xy)
2015 dst.zw = max(src0.zw, src1.zw)
2017 .. opcode:: DMUL - Multiply
2021 dst.xy = src0.xy \times src1.xy
2023 dst.zw = src0.zw \times src1.zw
2026 .. opcode:: DMAD - Multiply And Add
2030 dst.xy = src0.xy \times src1.xy + src2.xy
2032 dst.zw = src0.zw \times src1.zw + src2.zw
2035 .. opcode:: DRCP - Reciprocal
2039 dst.xy = \frac{1}{src.xy}
2041 dst.zw = \frac{1}{src.zw}
2043 .. opcode:: DSQRT - Square Root
2047 dst.xy = \sqrt{src.xy}
2049 dst.zw = \sqrt{src.zw}
2052 .. _samplingopcodes:
2054 Resource Sampling Opcodes
2055 ^^^^^^^^^^^^^^^^^^^^^^^^^
2057 Those opcodes follow very closely semantics of the respective Direct3D
2058 instructions. If in doubt double check Direct3D documentation.
2059 Note that the swizzle on SVIEW (src1) determines texel swizzling
2064 Using provided address, sample data from the specified texture using the
2065 filtering mode identified by the gven sampler. The source data may come from
2066 any resource type other than buffers.
2068 Syntax: ``SAMPLE dst, address, sampler_view, sampler``
2070 Example: ``SAMPLE TEMP[0], TEMP[1], SVIEW[0], SAMP[0]``
2072 .. opcode:: SAMPLE_I
2074 Simplified alternative to the SAMPLE instruction. Using the provided
2075 integer address, SAMPLE_I fetches data from the specified sampler view
2076 without any filtering. The source data may come from any resource type
2079 Syntax: ``SAMPLE_I dst, address, sampler_view``
2081 Example: ``SAMPLE_I TEMP[0], TEMP[1], SVIEW[0]``
2083 The 'address' is specified as unsigned integers. If the 'address' is out of
2084 range [0...(# texels - 1)] the result of the fetch is always 0 in all
2085 components. As such the instruction doesn't honor address wrap modes, in
2086 cases where that behavior is desirable 'SAMPLE' instruction should be used.
2087 address.w always provides an unsigned integer mipmap level. If the value is
2088 out of the range then the instruction always returns 0 in all components.
2089 address.yz are ignored for buffers and 1d textures. address.z is ignored
2090 for 1d texture arrays and 2d textures.
2092 For 1D texture arrays address.y provides the array index (also as unsigned
2093 integer). If the value is out of the range of available array indices
2094 [0... (array size - 1)] then the opcode always returns 0 in all components.
2095 For 2D texture arrays address.z provides the array index, otherwise it
2096 exhibits the same behavior as in the case for 1D texture arrays. The exact
2097 semantics of the source address are presented in the table below:
2099 +---------------------------+----+-----+-----+---------+
2100 | resource type | X | Y | Z | W |
2101 +===========================+====+=====+=====+=========+
2102 | ``PIPE_BUFFER`` | x | | | ignored |
2103 +---------------------------+----+-----+-----+---------+
2104 | ``PIPE_TEXTURE_1D`` | x | | | mpl |
2105 +---------------------------+----+-----+-----+---------+
2106 | ``PIPE_TEXTURE_2D`` | x | y | | mpl |
2107 +---------------------------+----+-----+-----+---------+
2108 | ``PIPE_TEXTURE_3D`` | x | y | z | mpl |
2109 +---------------------------+----+-----+-----+---------+
2110 | ``PIPE_TEXTURE_RECT`` | x | y | | mpl |
2111 +---------------------------+----+-----+-----+---------+
2112 | ``PIPE_TEXTURE_CUBE`` | not allowed as source |
2113 +---------------------------+----+-----+-----+---------+
2114 | ``PIPE_TEXTURE_1D_ARRAY`` | x | idx | | mpl |
2115 +---------------------------+----+-----+-----+---------+
2116 | ``PIPE_TEXTURE_2D_ARRAY`` | x | y | idx | mpl |
2117 +---------------------------+----+-----+-----+---------+
2119 Where 'mpl' is a mipmap level and 'idx' is the array index.
2121 .. opcode:: SAMPLE_I_MS
2123 Just like SAMPLE_I but allows fetch data from multi-sampled surfaces.
2125 Syntax: ``SAMPLE_I_MS dst, address, sampler_view, sample``
2127 .. opcode:: SAMPLE_B
2129 Just like the SAMPLE instruction with the exception that an additional bias
2130 is applied to the level of detail computed as part of the instruction
2133 Syntax: ``SAMPLE_B dst, address, sampler_view, sampler, lod_bias``
2135 Example: ``SAMPLE_B TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2].x``
2137 .. opcode:: SAMPLE_C
2139 Similar to the SAMPLE instruction but it performs a comparison filter. The
2140 operands to SAMPLE_C are identical to SAMPLE, except that there is an
2141 additional float32 operand, reference value, which must be a register with
2142 single-component, or a scalar literal. SAMPLE_C makes the hardware use the
2143 current samplers compare_func (in pipe_sampler_state) to compare reference
2144 value against the red component value for the surce resource at each texel
2145 that the currently configured texture filter covers based on the provided
2148 Syntax: ``SAMPLE_C dst, address, sampler_view.r, sampler, ref_value``
2150 Example: ``SAMPLE_C TEMP[0], TEMP[1], SVIEW[0].r, SAMP[0], TEMP[2].x``
2152 .. opcode:: SAMPLE_C_LZ
2154 Same as SAMPLE_C, but LOD is 0 and derivatives are ignored. The LZ stands
2157 Syntax: ``SAMPLE_C_LZ dst, address, sampler_view.r, sampler, ref_value``
2159 Example: ``SAMPLE_C_LZ TEMP[0], TEMP[1], SVIEW[0].r, SAMP[0], TEMP[2].x``
2162 .. opcode:: SAMPLE_D
2164 SAMPLE_D is identical to the SAMPLE opcode except that the derivatives for
2165 the source address in the x direction and the y direction are provided by
2168 Syntax: ``SAMPLE_D dst, address, sampler_view, sampler, der_x, der_y``
2170 Example: ``SAMPLE_D TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2], TEMP[3]``
2172 .. opcode:: SAMPLE_L
2174 SAMPLE_L is identical to the SAMPLE opcode except that the LOD is provided
2175 directly as a scalar value, representing no anisotropy.
2177 Syntax: ``SAMPLE_L dst, address, sampler_view, sampler, explicit_lod``
2179 Example: ``SAMPLE_L TEMP[0], TEMP[1], SVIEW[0], SAMP[0], TEMP[2].x``
2183 Gathers the four texels to be used in a bi-linear filtering operation and
2184 packs them into a single register. Only works with 2D, 2D array, cubemaps,
2185 and cubemaps arrays. For 2D textures, only the addressing modes of the
2186 sampler and the top level of any mip pyramid are used. Set W to zero. It
2187 behaves like the SAMPLE instruction, but a filtered sample is not
2188 generated. The four samples that contribute to filtering are placed into
2189 xyzw in counter-clockwise order, starting with the (u,v) texture coordinate
2190 delta at the following locations (-, +), (+, +), (+, -), (-, -), where the
2191 magnitude of the deltas are half a texel.
2194 .. opcode:: SVIEWINFO
2196 Query the dimensions of a given sampler view. dst receives width, height,
2197 depth or array size and number of mipmap levels as int4. The dst can have a
2198 writemask which will specify what info is the caller interested in.
2200 Syntax: ``SVIEWINFO dst, src_mip_level, sampler_view``
2202 Example: ``SVIEWINFO TEMP[0], TEMP[1].x, SVIEW[0]``
2204 src_mip_level is an unsigned integer scalar. If it's out of range then
2205 returns 0 for width, height and depth/array size but the total number of
2206 mipmap is still returned correctly for the given sampler view. The returned
2207 width, height and depth values are for the mipmap level selected by the
2208 src_mip_level and are in the number of texels. For 1d texture array width
2209 is in dst.x, array size is in dst.y and dst.z is 0. The number of mipmaps is
2210 still in dst.w. In contrast to d3d10 resinfo, there's no way in the tgsi
2211 instruction encoding to specify the return type (float/rcpfloat/uint), hence
2212 always using uint. Also, unlike the SAMPLE instructions, the swizzle on src1
2213 resinfo allowing swizzling dst values is ignored (due to the interaction
2214 with rcpfloat modifier which requires some swizzle handling in the state
2217 .. opcode:: SAMPLE_POS
2219 Query the position of a given sample. dst receives float4 (x, y, 0, 0)
2220 indicated where the sample is located. If the resource is not a multi-sample
2221 resource and not a render target, the result is 0.
2223 .. opcode:: SAMPLE_INFO
2225 dst receives number of samples in x. If the resource is not a multi-sample
2226 resource and not a render target, the result is 0.
2229 .. _resourceopcodes:
2231 Resource Access Opcodes
2232 ^^^^^^^^^^^^^^^^^^^^^^^
2234 .. opcode:: LOAD - Fetch data from a shader resource
2236 Syntax: ``LOAD dst, resource, address``
2238 Example: ``LOAD TEMP[0], RES[0], TEMP[1]``
2240 Using the provided integer address, LOAD fetches data
2241 from the specified buffer or texture without any
2244 The 'address' is specified as a vector of unsigned
2245 integers. If the 'address' is out of range the result
2248 Only the first mipmap level of a resource can be read
2249 from using this instruction.
2251 For 1D or 2D texture arrays, the array index is
2252 provided as an unsigned integer in address.y or
2253 address.z, respectively. address.yz are ignored for
2254 buffers and 1D textures. address.z is ignored for 1D
2255 texture arrays and 2D textures. address.w is always
2258 .. opcode:: STORE - Write data to a shader resource
2260 Syntax: ``STORE resource, address, src``
2262 Example: ``STORE RES[0], TEMP[0], TEMP[1]``
2264 Using the provided integer address, STORE writes data
2265 to the specified buffer or texture.
2267 The 'address' is specified as a vector of unsigned
2268 integers. If the 'address' is out of range the result
2271 Only the first mipmap level of a resource can be
2272 written to using this instruction.
2274 For 1D or 2D texture arrays, the array index is
2275 provided as an unsigned integer in address.y or
2276 address.z, respectively. address.yz are ignored for
2277 buffers and 1D textures. address.z is ignored for 1D
2278 texture arrays and 2D textures. address.w is always
2282 .. _threadsyncopcodes:
2284 Inter-thread synchronization opcodes
2285 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2287 These opcodes are intended for communication between threads running
2288 within the same compute grid. For now they're only valid in compute
2291 .. opcode:: MFENCE - Memory fence
2293 Syntax: ``MFENCE resource``
2295 Example: ``MFENCE RES[0]``
2297 This opcode forces strong ordering between any memory access
2298 operations that affect the specified resource. This means that
2299 previous loads and stores (and only those) will be performed and
2300 visible to other threads before the program execution continues.
2303 .. opcode:: LFENCE - Load memory fence
2305 Syntax: ``LFENCE resource``
2307 Example: ``LFENCE RES[0]``
2309 Similar to MFENCE, but it only affects the ordering of memory loads.
2312 .. opcode:: SFENCE - Store memory fence
2314 Syntax: ``SFENCE resource``
2316 Example: ``SFENCE RES[0]``
2318 Similar to MFENCE, but it only affects the ordering of memory stores.
2321 .. opcode:: BARRIER - Thread group barrier
2325 This opcode suspends the execution of the current thread until all
2326 the remaining threads in the working group reach the same point of
2327 the program. Results are unspecified if any of the remaining
2328 threads terminates or never reaches an executed BARRIER instruction.
2336 These opcodes provide atomic variants of some common arithmetic and
2337 logical operations. In this context atomicity means that another
2338 concurrent memory access operation that affects the same memory
2339 location is guaranteed to be performed strictly before or after the
2340 entire execution of the atomic operation.
2342 For the moment they're only valid in compute programs.
2344 .. opcode:: ATOMUADD - Atomic integer addition
2346 Syntax: ``ATOMUADD dst, resource, offset, src``
2348 Example: ``ATOMUADD TEMP[0], RES[0], TEMP[1], TEMP[2]``
2350 The following operation is performed atomically on each component:
2354 dst_i = resource[offset]_i
2356 resource[offset]_i = dst_i + src_i
2359 .. opcode:: ATOMXCHG - Atomic exchange
2361 Syntax: ``ATOMXCHG dst, resource, offset, src``
2363 Example: ``ATOMXCHG TEMP[0], RES[0], TEMP[1], TEMP[2]``
2365 The following operation is performed atomically on each component:
2369 dst_i = resource[offset]_i
2371 resource[offset]_i = src_i
2374 .. opcode:: ATOMCAS - Atomic compare-and-exchange
2376 Syntax: ``ATOMCAS dst, resource, offset, cmp, src``
2378 Example: ``ATOMCAS TEMP[0], RES[0], TEMP[1], TEMP[2], TEMP[3]``
2380 The following operation is performed atomically on each component:
2384 dst_i = resource[offset]_i
2386 resource[offset]_i = (dst_i == cmp_i ? src_i : dst_i)
2389 .. opcode:: ATOMAND - Atomic bitwise And
2391 Syntax: ``ATOMAND dst, resource, offset, src``
2393 Example: ``ATOMAND TEMP[0], RES[0], TEMP[1], TEMP[2]``
2395 The following operation is performed atomically on each component:
2399 dst_i = resource[offset]_i
2401 resource[offset]_i = dst_i \& src_i
2404 .. opcode:: ATOMOR - Atomic bitwise Or
2406 Syntax: ``ATOMOR dst, resource, offset, src``
2408 Example: ``ATOMOR TEMP[0], RES[0], TEMP[1], TEMP[2]``
2410 The following operation is performed atomically on each component:
2414 dst_i = resource[offset]_i
2416 resource[offset]_i = dst_i | src_i
2419 .. opcode:: ATOMXOR - Atomic bitwise Xor
2421 Syntax: ``ATOMXOR dst, resource, offset, src``
2423 Example: ``ATOMXOR TEMP[0], RES[0], TEMP[1], TEMP[2]``
2425 The following operation is performed atomically on each component:
2429 dst_i = resource[offset]_i
2431 resource[offset]_i = dst_i \oplus src_i
2434 .. opcode:: ATOMUMIN - Atomic unsigned minimum
2436 Syntax: ``ATOMUMIN dst, resource, offset, src``
2438 Example: ``ATOMUMIN TEMP[0], RES[0], TEMP[1], TEMP[2]``
2440 The following operation is performed atomically on each component:
2444 dst_i = resource[offset]_i
2446 resource[offset]_i = (dst_i < src_i ? dst_i : src_i)
2449 .. opcode:: ATOMUMAX - Atomic unsigned maximum
2451 Syntax: ``ATOMUMAX dst, resource, offset, src``
2453 Example: ``ATOMUMAX TEMP[0], RES[0], TEMP[1], TEMP[2]``
2455 The following operation is performed atomically on each component:
2459 dst_i = resource[offset]_i
2461 resource[offset]_i = (dst_i > src_i ? dst_i : src_i)
2464 .. opcode:: ATOMIMIN - Atomic signed minimum
2466 Syntax: ``ATOMIMIN dst, resource, offset, src``
2468 Example: ``ATOMIMIN TEMP[0], RES[0], TEMP[1], TEMP[2]``
2470 The following operation is performed atomically on each component:
2474 dst_i = resource[offset]_i
2476 resource[offset]_i = (dst_i < src_i ? dst_i : src_i)
2479 .. opcode:: ATOMIMAX - Atomic signed maximum
2481 Syntax: ``ATOMIMAX dst, resource, offset, src``
2483 Example: ``ATOMIMAX TEMP[0], RES[0], TEMP[1], TEMP[2]``
2485 The following operation is performed atomically on each component:
2489 dst_i = resource[offset]_i
2491 resource[offset]_i = (dst_i > src_i ? dst_i : src_i)
2495 Explanation of symbols used
2496 ------------------------------
2503 :math:`|x|` Absolute value of `x`.
2505 :math:`\lceil x \rceil` Ceiling of `x`.
2507 clamp(x,y,z) Clamp x between y and z.
2508 (x < y) ? y : (x > z) ? z : x
2510 :math:`\lfloor x\rfloor` Floor of `x`.
2512 :math:`\log_2{x}` Logarithm of `x`, base 2.
2514 max(x,y) Maximum of x and y.
2517 min(x,y) Minimum of x and y.
2520 partialx(x) Derivative of x relative to fragment's X.
2522 partialy(x) Derivative of x relative to fragment's Y.
2524 pop() Pop from stack.
2526 :math:`x^y` `x` to the power `y`.
2528 push(x) Push x on stack.
2532 trunc(x) Truncate x, i.e. drop the fraction bits.
2539 discard Discard fragment.
2543 target Label of target instruction.
2554 Declares a register that is will be referenced as an operand in Instruction
2557 File field contains register file that is being declared and is one
2560 UsageMask field specifies which of the register components can be accessed
2561 and is one of TGSI_WRITEMASK.
2563 The Local flag specifies that a given value isn't intended for
2564 subroutine parameter passing and, as a result, the implementation
2565 isn't required to give any guarantees of it being preserved across
2566 subroutine boundaries. As it's merely a compiler hint, the
2567 implementation is free to ignore it.
2569 If Dimension flag is set to 1, a Declaration Dimension token follows.
2571 If Semantic flag is set to 1, a Declaration Semantic token follows.
2573 If Interpolate flag is set to 1, a Declaration Interpolate token follows.
2575 If file is TGSI_FILE_RESOURCE, a Declaration Resource token follows.
2577 If Array flag is set to 1, a Declaration Array token follows.
2580 ^^^^^^^^^^^^^^^^^^^^^^^^
2582 Declarations can optional have an ArrayID attribute which can be referred by
2583 indirect addressing operands. An ArrayID of zero is reserved and treaded as
2584 if no ArrayID is specified.
2586 If an indirect addressing operand refers to a specific declaration by using
2587 an ArrayID only the registers in this declaration are guaranteed to be
2588 accessed, accessing any register outside this declaration results in undefined
2589 behavior. Note that for compatibility the effective index is zero-based and
2590 not relative to the specified declaration
2592 If no ArrayID is specified with an indirect addressing operand the whole
2593 register file might be accessed by this operand. This is strongly discouraged
2594 and will prevent packing of scalar/vec2 arrays and effective alias analysis.
2596 Declaration Semantic
2597 ^^^^^^^^^^^^^^^^^^^^^^^^
2599 Vertex and fragment shader input and output registers may be labeled
2600 with semantic information consisting of a name and index.
2602 Follows Declaration token if Semantic bit is set.
2604 Since its purpose is to link a shader with other stages of the pipeline,
2605 it is valid to follow only those Declaration tokens that declare a register
2606 either in INPUT or OUTPUT file.
2608 SemanticName field contains the semantic name of the register being declared.
2609 There is no default value.
2611 SemanticIndex is an optional subscript that can be used to distinguish
2612 different register declarations with the same semantic name. The default value
2615 The meanings of the individual semantic names are explained in the following
2618 TGSI_SEMANTIC_POSITION
2619 """"""""""""""""""""""
2621 For vertex shaders, TGSI_SEMANTIC_POSITION indicates the vertex shader
2622 output register which contains the homogeneous vertex position in the clip
2623 space coordinate system. After clipping, the X, Y and Z components of the
2624 vertex will be divided by the W value to get normalized device coordinates.
2626 For fragment shaders, TGSI_SEMANTIC_POSITION is used to indicate that
2627 fragment shader input contains the fragment's window position. The X
2628 component starts at zero and always increases from left to right.
2629 The Y component starts at zero and always increases but Y=0 may either
2630 indicate the top of the window or the bottom depending on the fragment
2631 coordinate origin convention (see TGSI_PROPERTY_FS_COORD_ORIGIN).
2632 The Z coordinate ranges from 0 to 1 to represent depth from the front
2633 to the back of the Z buffer. The W component contains the reciprocol
2634 of the interpolated vertex position W component.
2636 Fragment shaders may also declare an output register with
2637 TGSI_SEMANTIC_POSITION. Only the Z component is writable. This allows
2638 the fragment shader to change the fragment's Z position.
2645 For vertex shader outputs or fragment shader inputs/outputs, this
2646 label indicates that the resister contains an R,G,B,A color.
2648 Several shader inputs/outputs may contain colors so the semantic index
2649 is used to distinguish them. For example, color[0] may be the diffuse
2650 color while color[1] may be the specular color.
2652 This label is needed so that the flat/smooth shading can be applied
2653 to the right interpolants during rasterization.
2657 TGSI_SEMANTIC_BCOLOR
2658 """"""""""""""""""""
2660 Back-facing colors are only used for back-facing polygons, and are only valid
2661 in vertex shader outputs. After rasterization, all polygons are front-facing
2662 and COLOR and BCOLOR end up occupying the same slots in the fragment shader,
2663 so all BCOLORs effectively become regular COLORs in the fragment shader.
2669 Vertex shader inputs and outputs and fragment shader inputs may be
2670 labeled with TGSI_SEMANTIC_FOG to indicate that the register contains
2671 a fog coordinate. Typically, the fragment shader will use the fog coordinate
2672 to compute a fog blend factor which is used to blend the normal fragment color
2673 with a constant fog color. But fog coord really is just an ordinary vec4
2674 register like regular semantics.
2680 Vertex shader input and output registers may be labeled with
2681 TGIS_SEMANTIC_PSIZE to indicate that the register contains a point size
2682 in the form (S, 0, 0, 1). The point size controls the width or diameter
2683 of points for rasterization. This label cannot be used in fragment
2686 When using this semantic, be sure to set the appropriate state in the
2687 :ref:`rasterizer` first.
2690 TGSI_SEMANTIC_TEXCOORD
2691 """"""""""""""""""""""
2693 Only available if PIPE_CAP_TGSI_TEXCOORD is exposed !
2695 Vertex shader outputs and fragment shader inputs may be labeled with
2696 this semantic to make them replaceable by sprite coordinates via the
2697 sprite_coord_enable state in the :ref:`rasterizer`.
2698 The semantic index permitted with this semantic is limited to <= 7.
2700 If the driver does not support TEXCOORD, sprite coordinate replacement
2701 applies to inputs with the GENERIC semantic instead.
2703 The intended use case for this semantic is gl_TexCoord.
2706 TGSI_SEMANTIC_PCOORD
2707 """"""""""""""""""""
2709 Only available if PIPE_CAP_TGSI_TEXCOORD is exposed !
2711 Fragment shader inputs may be labeled with TGSI_SEMANTIC_PCOORD to indicate
2712 that the register contains sprite coordinates in the form (x, y, 0, 1), if
2713 the current primitive is a point and point sprites are enabled. Otherwise,
2714 the contents of the register are undefined.
2716 The intended use case for this semantic is gl_PointCoord.
2719 TGSI_SEMANTIC_GENERIC
2720 """""""""""""""""""""
2722 All vertex/fragment shader inputs/outputs not labeled with any other
2723 semantic label can be considered to be generic attributes. Typical
2724 uses of generic inputs/outputs are texcoords and user-defined values.
2727 TGSI_SEMANTIC_NORMAL
2728 """"""""""""""""""""
2730 Indicates that a vertex shader input is a normal vector. This is
2731 typically only used for legacy graphics APIs.
2737 This label applies to fragment shader inputs only and indicates that
2738 the register contains front/back-face information of the form (F, 0,
2739 0, 1). The first component will be positive when the fragment belongs
2740 to a front-facing polygon, and negative when the fragment belongs to a
2741 back-facing polygon.
2744 TGSI_SEMANTIC_EDGEFLAG
2745 """"""""""""""""""""""
2747 For vertex shaders, this sematic label indicates that an input or
2748 output is a boolean edge flag. The register layout is [F, x, x, x]
2749 where F is 0.0 or 1.0 and x = don't care. Normally, the vertex shader
2750 simply copies the edge flag input to the edgeflag output.
2752 Edge flags are used to control which lines or points are actually
2753 drawn when the polygon mode converts triangles/quads/polygons into
2757 TGSI_SEMANTIC_STENCIL
2758 """""""""""""""""""""
2760 For fragment shaders, this semantic label indicates that an output
2761 is a writable stencil reference value. Only the Y component is writable.
2762 This allows the fragment shader to change the fragments stencilref value.
2765 TGSI_SEMANTIC_VIEWPORT_INDEX
2766 """"""""""""""""""""""""""""
2768 For geometry shaders, this semantic label indicates that an output
2769 contains the index of the viewport (and scissor) to use.
2770 Only the X value is used.
2776 For geometry shaders, this semantic label indicates that an output
2777 contains the layer value to use for the color and depth/stencil surfaces.
2778 Only the X value is used. (Also known as rendertarget array index.)
2781 TGSI_SEMANTIC_CULLDIST
2782 """"""""""""""""""""""
2784 Used as distance to plane for performing application-defined culling
2785 of individual primitives against a plane. When components of vertex
2786 elements are given this label, these values are assumed to be a
2787 float32 signed distance to a plane. Primitives will be completely
2788 discarded if the plane distance for all of the vertices in the
2789 primitive are < 0. If a vertex has a cull distance of NaN, that
2790 vertex counts as "out" (as if its < 0);
2791 The limits on both clip and cull distances are bound
2792 by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_COUNT define which defines
2793 the maximum number of components that can be used to hold the
2794 distances and by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_ELEMENT_COUNT
2795 which specifies the maximum number of registers which can be
2796 annotated with those semantics.
2799 TGSI_SEMANTIC_CLIPDIST
2800 """"""""""""""""""""""
2802 When components of vertex elements are identified this way, these
2803 values are each assumed to be a float32 signed distance to a plane.
2804 Primitive setup only invokes rasterization on pixels for which
2805 the interpolated plane distances are >= 0. Multiple clip planes
2806 can be implemented simultaneously, by annotating multiple
2807 components of one or more vertex elements with the above specified
2808 semantic. The limits on both clip and cull distances are bound
2809 by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_COUNT define which defines
2810 the maximum number of components that can be used to hold the
2811 distances and by the PIPE_MAX_CLIP_OR_CULL_DISTANCE_ELEMENT_COUNT
2812 which specifies the maximum number of registers which can be
2813 annotated with those semantics.
2815 TGSI_SEMANTIC_SAMPLEID
2816 """"""""""""""""""""""
2818 For fragment shaders, this semantic label indicates that a system value
2819 contains the current sample id (i.e. gl_SampleID). Only the X value is used.
2821 TGSI_SEMANTIC_SAMPLEPOS
2822 """""""""""""""""""""""
2824 For fragment shaders, this semantic label indicates that a system value
2825 contains the current sample's position (i.e. gl_SamplePosition). Only the X
2826 and Y values are used.
2828 TGSI_SEMANTIC_SAMPLEMASK
2829 """"""""""""""""""""""""
2831 For fragment shaders, this semantic label indicates that an output contains
2832 the sample mask used to disable further sample processing
2833 (i.e. gl_SampleMask). Only the X value is used, up to 32x MS.
2835 TGSI_SEMANTIC_INVOCATIONID
2836 """"""""""""""""""""""""""
2838 For geometry shaders, this semantic label indicates that a system value
2839 contains the current invocation id (i.e. gl_InvocationID). Only the X value is
2842 Declaration Interpolate
2843 ^^^^^^^^^^^^^^^^^^^^^^^
2845 This token is only valid for fragment shader INPUT declarations.
2847 The Interpolate field specifes the way input is being interpolated by
2848 the rasteriser and is one of TGSI_INTERPOLATE_*.
2850 The Location field specifies the location inside the pixel that the
2851 interpolation should be done at, one of ``TGSI_INTERPOLATE_LOC_*``. Note that
2852 when per-sample shading is enabled, the implementation may choose to
2853 interpolate at the sample irrespective of the Location field.
2855 The CylindricalWrap bitfield specifies which register components
2856 should be subject to cylindrical wrapping when interpolating by the
2857 rasteriser. If TGSI_CYLINDRICAL_WRAP_X is set to 1, the X component
2858 should be interpolated according to cylindrical wrapping rules.
2861 Declaration Sampler View
2862 ^^^^^^^^^^^^^^^^^^^^^^^^
2864 Follows Declaration token if file is TGSI_FILE_SAMPLER_VIEW.
2866 DCL SVIEW[#], resource, type(s)
2868 Declares a shader input sampler view and assigns it to a SVIEW[#]
2871 resource can be one of BUFFER, 1D, 2D, 3D, 1DArray and 2DArray.
2873 type must be 1 or 4 entries (if specifying on a per-component
2874 level) out of UNORM, SNORM, SINT, UINT and FLOAT.
2877 Declaration Resource
2878 ^^^^^^^^^^^^^^^^^^^^
2880 Follows Declaration token if file is TGSI_FILE_RESOURCE.
2882 DCL RES[#], resource [, WR] [, RAW]
2884 Declares a shader input resource and assigns it to a RES[#]
2887 resource can be one of BUFFER, 1D, 2D, 3D, CUBE, 1DArray and
2890 If the RAW keyword is not specified, the texture data will be
2891 subject to conversion, swizzling and scaling as required to yield
2892 the specified data type from the physical data format of the bound
2895 If the RAW keyword is specified, no channel conversion will be
2896 performed: the values read for each of the channels (X,Y,Z,W) will
2897 correspond to consecutive words in the same order and format
2898 they're found in memory. No element-to-address conversion will be
2899 performed either: the value of the provided X coordinate will be
2900 interpreted in byte units instead of texel units. The result of
2901 accessing a misaligned address is undefined.
2903 Usage of the STORE opcode is only allowed if the WR (writable) flag
2908 ^^^^^^^^^^^^^^^^^^^^^^^^
2910 Properties are general directives that apply to the whole TGSI program.
2915 Specifies the fragment shader TGSI_SEMANTIC_POSITION coordinate origin.
2916 The default value is UPPER_LEFT.
2918 If UPPER_LEFT, the position will be (0,0) at the upper left corner and
2919 increase downward and rightward.
2920 If LOWER_LEFT, the position will be (0,0) at the lower left corner and
2921 increase upward and rightward.
2923 OpenGL defaults to LOWER_LEFT, and is configurable with the
2924 GL_ARB_fragment_coord_conventions extension.
2926 DirectX 9/10 use UPPER_LEFT.
2928 FS_COORD_PIXEL_CENTER
2929 """""""""""""""""""""
2931 Specifies the fragment shader TGSI_SEMANTIC_POSITION pixel center convention.
2932 The default value is HALF_INTEGER.
2934 If HALF_INTEGER, the fractionary part of the position will be 0.5
2935 If INTEGER, the fractionary part of the position will be 0.0
2937 Note that this does not affect the set of fragments generated by
2938 rasterization, which is instead controlled by half_pixel_center in the
2941 OpenGL defaults to HALF_INTEGER, and is configurable with the
2942 GL_ARB_fragment_coord_conventions extension.
2944 DirectX 9 uses INTEGER.
2945 DirectX 10 uses HALF_INTEGER.
2947 FS_COLOR0_WRITES_ALL_CBUFS
2948 """"""""""""""""""""""""""
2949 Specifies that writes to the fragment shader color 0 are replicated to all
2950 bound cbufs. This facilitates OpenGL's fragColor output vs fragData[0] where
2951 fragData is directed to a single color buffer, but fragColor is broadcast.
2954 """"""""""""""""""""""""""
2955 If this property is set on the program bound to the shader stage before the
2956 fragment shader, user clip planes should have no effect (be disabled) even if
2957 that shader does not write to any clip distance outputs and the rasterizer's
2958 clip_plane_enable is non-zero.
2959 This property is only supported by drivers that also support shader clip
2961 This is useful for APIs that don't have UCPs and where clip distances written
2962 by a shader cannot be disabled.
2967 Specifies the number of times a geometry shader should be executed for each
2968 input primitive. Each invocation will have a different
2969 TGSI_SEMANTIC_INVOCATIONID system value set. If not specified, assumed to
2972 VS_WINDOW_SPACE_POSITION
2973 """"""""""""""""""""""""""
2974 If this property is set on the vertex shader, the TGSI_SEMANTIC_POSITION output
2975 is assumed to contain window space coordinates.
2976 Division of X,Y,Z by W and the viewport transformation are disabled, and 1/W is
2977 directly taken from the 4-th component of the shader output.
2978 Naturally, clipping is not performed on window coordinates either.
2979 The effect of this property is undefined if a geometry or tessellation shader
2982 Texture Sampling and Texture Formats
2983 ------------------------------------
2985 This table shows how texture image components are returned as (x,y,z,w) tuples
2986 by TGSI texture instructions, such as :opcode:`TEX`, :opcode:`TXD`, and
2987 :opcode:`TXP`. For reference, OpenGL and Direct3D conventions are shown as
2990 +--------------------+--------------+--------------------+--------------+
2991 | Texture Components | Gallium | OpenGL | Direct3D 9 |
2992 +====================+==============+====================+==============+
2993 | R | (r, 0, 0, 1) | (r, 0, 0, 1) | (r, 1, 1, 1) |
2994 +--------------------+--------------+--------------------+--------------+
2995 | RG | (r, g, 0, 1) | (r, g, 0, 1) | (r, g, 1, 1) |
2996 +--------------------+--------------+--------------------+--------------+
2997 | RGB | (r, g, b, 1) | (r, g, b, 1) | (r, g, b, 1) |
2998 +--------------------+--------------+--------------------+--------------+
2999 | RGBA | (r, g, b, a) | (r, g, b, a) | (r, g, b, a) |
3000 +--------------------+--------------+--------------------+--------------+
3001 | A | (0, 0, 0, a) | (0, 0, 0, a) | (0, 0, 0, a) |
3002 +--------------------+--------------+--------------------+--------------+
3003 | L | (l, l, l, 1) | (l, l, l, 1) | (l, l, l, 1) |
3004 +--------------------+--------------+--------------------+--------------+
3005 | LA | (l, l, l, a) | (l, l, l, a) | (l, l, l, a) |
3006 +--------------------+--------------+--------------------+--------------+
3007 | I | (i, i, i, i) | (i, i, i, i) | N/A |
3008 +--------------------+--------------+--------------------+--------------+
3009 | UV | XXX TBD | (0, 0, 0, 1) | (u, v, 1, 1) |
3010 | | | [#envmap-bumpmap]_ | |
3011 +--------------------+--------------+--------------------+--------------+
3012 | Z | XXX TBD | (z, z, z, 1) | (0, z, 0, 1) |
3013 | | | [#depth-tex-mode]_ | |
3014 +--------------------+--------------+--------------------+--------------+
3015 | S | (s, s, s, s) | unknown | unknown |
3016 +--------------------+--------------+--------------------+--------------+
3018 .. [#envmap-bumpmap] http://www.opengl.org/registry/specs/ATI/envmap_bumpmap.txt
3019 .. [#depth-tex-mode] the default is (z, z, z, 1) but may also be (0, 0, 0, z)
3020 or (z, z, z, z) depending on the value of GL_DEPTH_TEXTURE_MODE.